options.py

import os
import argparse

import numpy as np
import torch


class Options():
    """Base option class"""

    def __init__(self):
        self.parser = argparse.ArgumentParser()
        self.parser.add_argument('--run_dir', default='_scratch/test_run', type=str, help='experiment directory')
        self.parser.add_argument('--dataset', default='clevr', type=str, help='select dataset, options: clevr, clevr-humans')
        # Dataloader
        self.parser.add_argument('--shuffle', default=1, type=int, help='shuffle dataset')
        self.parser.add_argument('--gpu_ids', default='0', type=str, help='ids of gpu to be used')
        self.parser.add_argument('--num_workers', default=1, type=int, help='number of workers for loading data')
        # Run
        self.parser.add_argument('--manual_seed', default=None, type=int, help='manual seed')
        # Dataset catalog
        # - CLEVR
        self.parser.add_argument('--clevr_train_scene_path', default='../data/raw/CLEVR_v1.0/scenes/CLEVR_train_scenes.json',
                                 type=str, help='path to clevr train scenes')
        self.parser.add_argument('--clevr_val_scene_path', default='../data/raw/CLEVR_v1.0/scenes/CLEVR_val_scenes.json',
                                 type=str, help='path to clevr val scenes')
        self.parser.add_argument('--clevr_train_question_path', default='../data/reason/clevr_h5/clevr_train_questions.h5',
                                 type=str, help='path to clevr train questions')
        self.parser.add_argument('--clevr_val_question_path', default='../data/reason/clevr_h5/clevr_val_questions.h5',
                                 type=str, help='path to clevr val questions')
        
        self.parser.add_argument('--clevr_train_image_path', default='',
                                 type=str, help='path to clevr train images')
        self.parser.add_argument('--clevr_val_image_path', default='',
                                 type=str, help='path to clevr val images')
        
        self.parser.add_argument('--clevr_vocab_path', default='../data/reason/clevr_h5/clevr_vocab.json',
                                 type=str, help='path to clevr vocab')
        # Weights
        self.parser.add_argument('--language_weights', default='',
                            type=str, help='language weights')
        self.parser.add_argument('--vision_weights', default='',
                    type=str, help='vision weights')
        self.parser.add_argument('--theory', default='',
            type=str, help='ASP theory program path')
        self.parser.add_argument('--abduction', default='',
            type=str, help='ASP abduction program path')
        
        
        self.parser.add_argument('--ground_truth', default=False,
            type=bool, help='Use ground truth')
        
        self.parser.add_argument('--load_checkpoint_path', required=True, type=str, help='checkpoint path')
        # self.parser.add_argument('--save_result_path', required=True, type=str, help='save result path')
        self.parser.add_argument('--max_val_samples', default=None, type=int, help='max val data')
        self.parser.add_argument('--batch_size', default=10, type=int, help='batch_size')

        self.is_train = False
        
        

    def parse(self):
        # Instantiate option
        self.opt = self.parser.parse_args()

        # Parse gpu id list
        str_gpu_ids = self.opt.gpu_ids.split(',')
        self.opt.gpu_ids = []
        for str_id in str_gpu_ids:
            if str_id.isdigit() and int(str_id) >= 0:
                self.opt.gpu_ids.append(int(str_id))
        if len(self.opt.gpu_ids) > 0 and torch.cuda.is_available():
            torch.cuda.set_device(self.opt.gpu_ids[0])
        else:
            print('| using cpu')
            self.opt.gpu_ids = []

        # Set manual seed
        if self.opt.manual_seed is not None:
            torch.manual_seed(self.opt.manual_seed)
            if len(self.opt.gpu_ids) > 0 and torch.cuda.is_available():
                torch.cuda.manual_seed(self.opt.manual_seed)

        # Print and save options
        args = vars(self.opt)
        # print('| options')
        # for k, v in args.items():
        #     print('%s: %s' % (str(k), str(v)))
        if not os.path.isdir(self.opt.run_dir):
            os.makedirs(self.opt.run_dir)
        if self.is_train:
            file_path = os.path.join(self.opt.run_dir, 'train_opt.txt')
        else:
            file_path = os.path.join(self.opt.run_dir, 'test_opt.txt')
        with open(file_path, 'wt') as fout:
            fout.write('| options\n')
            for k, v in args.items():
                fout.write('%s: %s\n' % (str(k), str(v)))

        return self.opt